ARTICLE IN PRESS Molecular Aspects of Medicine

Contents lists available at ScienceDirect

Molecular Aspects of Medicine

j o u r n a l h o m e p a g e : w w w. e l s e v i e r. c o m / l o c a t e / m a m

Review

Inflammation, vitamin B6 and related pathways

c c c

Per Magne Ueland

*, Adrian McCann , Øivind Midttun , Arve Ulvik

a Department of Clinical Science, University of Bergen, 5021 Bergen, Norway

b Laboratory of Clinical Biochemistry, Haukeland University Hospital, 5021 Bergen, Norway

c Bevital A/S, Laboratoriebygget, 5021 Bergen, Norway

A R T I C L E I N F O A B S T R A C T

Article history:

Received 24 May 2016 Accepted 27 August 2016 Available online

Keywords:

Vitamin B6 Inflammation Kynurenines

Sphingosine 1-phosphate Transsulfuration pathway Glycine cleavage system (GCS)

Serine hydroxymethyltransferase (SHMT)

The active form of vitamin B6, pyridoxal 5′-phosphate (PLP), serves as a co-factor in more than 150 enzymatic reactions. Plasma PLP has consistently been shown to be low in inflammatory conditions; there is a parallel reduction in liver PLP, but minor changes in erythrocyte and muscle PLP and in functional vitamin B6 biomarkers. Plasma PLP also predicts the risk of chronic dis­

eases like cardiovascular disease and some cancers, and is inversely associated with numerous inflammatory markers in clinical and population-based studies. Vitamin B6 intake and supple­

mentation improve some immune functions in vitamin B6-deficient humans and experimental animals. A possible mechanism involved is mobilization of vitamin B6 to the sites of inflam­

mation where it may serve as a co-factor in pathways producing metabolites with immunomodulating effects. Relevant vitamin B6-dependent inflammatory pathways include vitamin B6 catabolism, the kynurenine pathway, sphingosine 1-phosphate metabolism, the transsulfuration pathway, and serine and glycine metabolism.

© 2016 Elsevier Ltd. All rights reserved.

Contents

1. Introduction ... 2

2. Vitamin B6 status and diseases ... 3

3. Vitamin B6, immune function and inflammation ... 3

4. Vitamin B6 and markers of inflammation ... 4

5. B6 catabolism and the PAr index ... 4

6. Mobilization of vitamin B6 during inflammation ... 5

7. Vitamin B6 and inflammatory pathways ... 5

7.1. The kynurenine pathway ... 5

7.1.1. The pathway ... 5

7.1.2. Key enzymes ... 7

7.1.3. Biological effects from metabolites ... 7

7.1.4. Kynurenine pathway metabolites and vitamin B6 status ... 8

7.1.5. Kynurenine pathway metabolites and chronic diseases ... 8

7.2. Sphingolipids ... 9

7.3. Transsulfuration pathway and hydrogen sulfide formation ... 9

7.4. Serine and glycine ... 12

8. Conclusion ... 13

References ... 14

* Corresponding author. Department of Clinical Science, University of Bergen, 5021 Bergen, Norway.

E-mail address: [email protected] (P.M. Ueland).

http://dx.doi.org/10.1016/j.mam.2016.08.001 0098-2997/© 2016 Elsevier Ltd. All rights reserved.

1. Introduction

Animal products mostly contain PLP and PMP, whereas in plant-derived products, PN(P) is the prevailing/principal B6 Vitamin B6 is a generic name that includes three form. Ingested PLP, PMP and PNP are dephosphorylated by different pyridine derivatives modified at their 4-position the ecto-enzyme tissue-specific intestinal phosphatase, prior and denoted pyridoxal (PL), pyridoxamine (PM) and pyri- to absorption. The portal circulation delivers PL, PM and PN doxine (PN), carrying an aldehyde, aminomethyl and to the liver, where they are rephosphorylated by pyridoxal hydroxymethyl group, respectively. All three forms exist as kinase (PDXK), and PMP and PNP are converted to PLP in derivatives that are phosphorylated at the 5-position. These reactions catalyzed by pyridoxine (pyridoxamine) oxidase are pyridoxal 5′-phosphate (PLP), pyridoxamine 5′-phosphate (PNPO) (Albersen et al., 2013; Coburn, 2015). In the liver, (PMP) and pyridoxine 5′-phosphate (PNP) (Coburn, 1996). PLP production is regulated (Merrill et al., 1978) such that The structure and metabolism of B6 vitamers are summa- the content remains relatively constant, even after a very rized in Fig. 1. high intake of PN (Schaeffer et al., 1989). The liver is also PLP is not synthesized

the production site of PLP destined for release into plasma tained from various foods including meat, milk products, (Huang et al., 2012; Lumeng et al., 1974), which in addi­

beans, nuts, potatoes and several fruits and vegetables. tion to PLP (about 70–80%) contains PL (8–30%) and the

HO

OH

O H

OH

AT

H₂N

OH

HO O P O

PNPO

PNPO

HO O P O

OH OH OH

H₃C N H₃C N H₃C N

Pyridoxine 5´-phosphate (PNP) Pyridoxal 5´-phosphate (PLP) Pyridoxamine 5´-phosphate (PMP)

PDXP

PDXP PDXK PDXK PDXP PDXK

ALP

HO O H H₂N

HO OH HO OH HO OH

C N

H₃C N H₃C N H₃

Pyridoxine (PN) Pyridoxal (PL) Pyridoxamine (PM)

AOX

O OH

HO OH

H₃C N

4-Pyridoxic acid (PA)

Fig. 1. Vitamin B6 forms, metabolism and the enzymes involved. Pyridoxal 5′-phosphate (PLP) is the metabolically active co-enzyme form of vitamin B6, pyridoxal (PL) is the transport form that crosses biological membranes, pyridoxine (PN) the vitamin B6 species present in supplements, and 4-pyridoxic acid (PA) is the vitamin B6 catabolite. PL, PN, and pyridoxamine (PM) are phosphorylated by pyridoxal (pyridoxine, vitamin B6) kinase (PDXK; EC 2.7.1.3511) to PLP, pyridoxine 5′-phosphate (PNP), and pyridoxamine 5′-phosphate (PMP), respectively. PNP and PMP are oxidized to PLP in reactions catalyzed by pyridoxamine 5′-phosphate oxidase (PNPO; EC 1.4.3.5). PLP in the liver is bound to albumin and is exported into the circulation for delivery to tissues.

Before cellular uptake, extracellular PLP is dephosphorylated to PL by the ectoenzyme, tissue-nonspecific alkaline phosphatase (ALP; EC 3.1.3.1). Dephos­

phorylation of PLP to PL is also catalyzed by the intracellular enzyme, pyridoxal (pyridoxine, vitamin B6) phosphatase (pyridoxal phosphate phosphatase, PDXP; EC 3.1.3.74). The irreversible oxidation of PL to PA is catalyzed by aldehyde oxidase(s) (AOX). Abbreviation: AT, aminotransferase. Modified from Ueland et al. (2015).

vitamin B6 catabolite, 4-pyridoxic acid (PA) (Leklem, 1990;

Talwar et al., 2003). PLP is exported from the liver bound to albumin. In plasma, PLP remains tightly bound to albumin (Huang et al., 2012; Lumeng et al., 1974) and increases over a broad range of vitamin B6 intake (Hansen et al., 1997, 2001). Plasma PLP must be dephosphorylated to the trans­

port form, PL, before being taken up by tissues or cells or passing the blood–brain barrier. The dephosphorylation is catalyzed by tissue-specific phosphatases, expressed in pla­

centa and germ cells, and the tissue-nonspecific alkaline phosphatase (ALP), which is an ectoenzyme located on the outer membrane of cells, including erythrocytes (Buchet et al., 2013).

PLP, the active B6 vitamer, serves as co-factor for more than 150 enzymes, which constitute about 4% of all enzyme activities (Percudani and Peracchi, 2009). These enzymes catalyze a wide range of reactions involving amino acids and amines, including transaminations, aldol cleavages, α-decarboxylations, racemizations, β- and γ-eliminations, and replacement reactions (Eliot and Kirsch, 2004). Most re­

actions are part of amino acid synthesis and degradation, while others are related to one-carbon metabolism, lipid me­

tabolism, gluconeogenesis, heme and neurotransmitter biosynthesis (Eliot and Kirsch, 2004; Percudani and Peracchi, 2009). As well as functioning as a co-factor, vitamin B6 has been described as a scavenger of reactive oxygen species (Kannan and Jain, 2004), metal chelator (Wondrak and Jacobson, 2012) and chaperone in the enzyme folding process (Cellini et al., 2014).

2. Vitamin B6 status and diseases

Isolated dietary vitamin B6 deficiency is rare in devel­

oped countries, but low circulating vitamin B6 has been reported in users of oral contraceptives or some drugs (Spinneker et al., 2007), in smokers (Ulvik et al., 2010), and in subjects with alcoholism (Ulvik et al., 2010), coeliac disease or diabetes (da Silva et al., 2012). Low vitamin B6 intake is associated with increased risk of cardiovascular disease (Rimm et al., 1998; Tavani et al., 2004) and cancer in some (Lim et al., 2005; Theodoratou et al., 2008; Wei et al., 2005) but not all (Larsson et al., 2010; Zhang et al., 2013) prospective studies. In general, the observed associations between low circulating PLP and risk of chronic diseases are generally more consistent (Page et al., 2009; Zhang et al., 2013), which may reflect a rather weak relation between vitamin B6 intake and plasma PLP (Larsson et al., 2010;

Rimm et al., 1998).

Low plasma PLP has been associated with risk of car­

diovascular disease (Cheng et al., 2008; Dalery et al., 1995;

Folsom et al., 1998; Friso et al., 2004; Page et al., 2009;

Robinson et al., 1998; Vanuzzo et al., 2007; Verhoef et al., 1996), stroke (Kelly et al., 2003, 2004) and venous throm­

bosis (Hron et al., 2007) in several, including three prospective studies (Folsom et al., 1998; Page et al., 2009;

Vanuzzo et al., 2007). However, it has been conjectured that systemic inflammation, as measured by elevated C-reactive protein (CRP), accounts for the prediction of myocardial in­

farction by plasma PLP (Dierkes et al., 2007). Prospective associations have also been observed between low plasma PLP and several cancers, including cancer of the ventricle

(Eussen et al., 2010), colorectum (Larsson et al., 2010; Le Marchand et al., 2009; Lee et al., 2009), lung (Johansson et al., 2010), breast (Lurie et al., 2012; Wu et al., 2013) and kidney (Johansson et al., 2014). In patients with established kidney cancer, high plasma PLP was associated with lower mor­

tality (Muller et al., 2015). Low plasma PLP has also been linked to rheumatoid arthritis (RA) (Chiang et al., 2005;

Huang et al., 2010; Roubenoff et al., 1995), inflammatory bowel disease (IBD) (Selhub et al., 2013), and diabetes (Friedman et al., 2004). As with cardiovascular disease and cancer, these are all conditions where inflammation is be­

lieved to play a key role in pathogenesis or disease progression (de Visser et al., 2006; Hansson et al., 2006; Tan et al., 2010).

In patients with IBD (Saibeni et al., 2003) or RA (Chiang et al., 2003a, 2003b), plasma PLP was inversely associated with the severity of the disease. RA patients had normal erythrocyte PLP but low plasma PLP (Chiang et al., 2005), that was not explained by low B6 intake, congenital defects in vitamin B6 metabolism (Chiang et al., 2003b; Paul et al., 2013) or vitamin B6 deficiency, as judged by a panel of func­

tional vitamin B6 biomarkers (Chiang et al., 2003a, 2003b;

Roubenoff et al., 1995). An experimental study in rats with adjuvant arthritis showed that affected animals had low plasma PLP, which correlated with low liver PLP, whereas the PLP content in muscle, the major PLP pool, was not af­

fected (Chiang et al., 2005). Furthermore, urinary PA excretion was not increased in RA patients and rats with ad­

juvant arthritis, suggesting no excessive vitamin B6 catabolism (Chiang et al., 2005).

Vitamin B6 status in critically ill patients as assessed by a transaminase activation assay and vitamin B6 intake were similar to these parameters in healthy controls (Huang et al., 2005), but the patients had reduced plasma PLP, elevated plasma PA (Huang et al., 2005), and normal (Quasim et al., 2005) or slightly reduced (Vasilaki et al., 2008) erythro­

cyte PLP; both plasma PLP and PA were associated with indices of immune response (Huang et al., 2005). Similar­

ly, in patients with myocardial infarction, plasma PLP showed a transient 40% fall with a nadir at about 40 hours after ad­

mission; the decline was accompanied by an equivalent increase in erythrocyte PLP. Plasma PLP returned to normal levels whereas erythrocyte PLP stayed elevated at the time of discharge (Vermaak et al., 1988). Supplementation of crit­

ically ill patients who had systemic inflammation with high dose pyridoxine caused no (Quasim et al., 2005) or a slight increase in plasma PLP (Cheng et al., 2006), a moderate (3­

fold) increase in erythrocyte PLP (Quasim et al., 2005), and a drastic (15–20 fold) increase in plasma PL (Cheng et al., 2006).

3. Vitamin B6, immune function and inflammation

Vitamin B6 deficiency affects cell-mediated immunity and to a lesser extent humoral immunity in both animal and human studies (Chandra and Sudhakaran, 1990; Rall and Meydani, 1993). A profound reduction in lymphocyte proliferation, T-cell mediated cytotoxicity, delayed-type hy­

persensitivity, allograft rejection (Rall and Meydani,

1993) and altered cytokine profile (Doke et al., 1998) have

been demonstrated in experimental studies on vitamin

B6-deficient rodents. Immune response in elderly sub­

jects (Talbott et al., 1987), patients with renal failure (Casciato et al., 1984), and critically ill patients (Cheng et al., 2006) is improved by supplementation with pyridoxine. The immune responses in elderly (Meydani et al., 1991), and in young women (Kwak et al., 2002) have been investigated in well-controlled metabolic settings as a function of vari­

able vitamin B6 status. In the elderly, vitamin B6 depletion decreased the number and mitogen response of blood lym­

phocytes, in particular T-helper cells, and interleukin (IL) 2 production. The immune indices were normalized upon vitamin B6 repletion (Meydani et al., 1991). In young women consuming 1 mg vitamin B6 per day (slightly below the rec­

ommended RDA of 1.3 mg/day) for one week, lymphocyte proliferation (but not IL-2 production) increased in a dose­

dependent manner as a function of vitamin B6 intake up to 2.1 mg/day. This suggests that vitamin B6 intake higher than the current RDA is required for maximum

phocyte mitogen response (Kwak et al., 2002).

4. Vitamin B6 and markers of inflammation

Plasma PLP shows an inverse association with inflam­

matory markers in clinical (Friedman et al., 2004; Friso et al., 2004; Huang et al., 2005; Saibeni et al., 2003; Ulvik et al., 2012) and population-based studies. Studies on population­

based cohorts have demonstrated that plasma PLP is inversely related to numerous markers of inflammation, in­

cluding CRP (Friso et al., 2001; Morris et al., 2010; Shen et al., 2010), IL-6 receptor (Gori et al., 2006), α-1­

antichymotrypsin (Bates et al., 1999a, 1999b), serum amyloid A (Abbenhardt et al., 2014), white blood cell count (WBC), kynurenine/tryptophan ratio (KTR), neopterin (Midttun et al., 2011; Theofylaktopoulou et al., 2014), and to an overall inflammatory summary score and summary scores repre­

senting different inflammatory modalities (Sakakeeny et al., 2012).

In the population based NHANES study (Morris et al., 2010), vitamin B6 intake from diet and supplements was inversely associated with CRP. Higher intake but also ade­

quate plasma PLP (

20 nmol/L) independent of intake appeared to protect against inflammation. Among sub­

jects with vitamin B6 intake of 2–3 mg/d, vitamin B6 inadequacy (plasma PLP

20 nmol/L) was uncommon (10%) in individuals with low CRP (≤3 mg/ML) but occurred more frequently (50%) in individuals with high CRP (

10 mg/L).

These results could be obtained both if vitamin B6 pro­

tects against inflammation and if inflammation adversely affects vitamin B6 status as measured by plasma PLP (Morris et al., 2010). Notably, in a small study on healthy individu­

als, controlled dietary vitamin B6 restriction did not affect the CRP levels (Davis et al., 2006), and in patients with stable angina pectoris, levels of inflammatory markers like CRP, neopterin or soluble CD40 ligand were not changed follow­

ing supplementation with high-dose (40 mg/d) PN alone or in combination with folic acid and vitamin B12 (Bleie et al., 2007). A strong inverse association between CRP and plasma PLP was maintained in cardiovascular patients even after supplementation (Ulvik et al., 2012). Thus, intervention studies suggest that the inverse association between CRP and plasma PLP reflects altered vitamin B6 distribution

during inflammation rather than high B6 protecting against inflammatory reactions.

Plasma PLP was inversely associated with CRP, WBC, KTR and neopterin, whereas PA showed a positive association with neopterin and KTR in patients with stable angina pec­

toris. These associations were essentially upheld after supplementation with high dose pyridoxine for 28 days. After supplementation, all B6 vitamers were increased 9–60-fold, but there was a steep drop in PL and PA in subjects with

CRP

7 mg/L (Ulvik et al., 2012). These data suggest that

acute-phase reaction (reflected by elevated CRP) leads to in­

creased uptake of vitamin B6 into tissues whereas cellular Th1 immune activation (neopterin and KTR) promoted uptake and a concurrent catabolism to PA.

5. B6 catabolism and the PAr index

4-Pyridoxic acid (PA) is a vitamin B6 catabolite formed in the liver from PL (Merrill et al., 1984). PA in plasma is increased after vitamin B6 intake (Bates et al., 1999a, 1999b;

Hansen et al., 2001), is not protein-bound (Anderson et al., 1974), has high renal clearance and is excreted in the urine (Coburn et al., 2002; Zempleni and Kübler, 1995). PA, as opposed to PLP, is not related to acute-phase inflammato­

ry status in the general population (Bates et al., 1999a, 1999b), but is positively related to markers of cellular immune activation (Ulvik et al., 2012) and is markedly in­

creased in critically ill patients (Huang et al., 2005).

The major circulating B6 vitamers, PLP, PL and PA, are measured simultaneously by contemporary methods based on mass spectrometry (Midttun et al., 2009), and demon­

strate a strong intercorrelation (Bor et al., 2003; Midttun et al., 2007), which may reflect tight metabolic control. The ratio PA/(PLP

PL), termed PAr index, was selected from other possible B6 vitamer combinations based on its high within-subject reproducibility (ICC of 0.75) (Ulvik et al., 2014), which suggests that PAr reflects key processes related to an individual’s vitamin B6 homeostasis.

PAr has some unique characteristics. Compared to the isolated B6 vitamers in plasma, PAr is less influenced by renal function, smoking, and vitamin B6 intake (Ulvik et al., 2014).

Approximately 90% of the explained variance of PAr is ac­

counted for by a summary score that includes four inflammatory markers (CRP, WBC, KTR and neopterin), and PAr efficiently discriminates high inflammatory status in ROC analyses (AUC of 0.85) (Ulvik et al., 2014).

The PAr response can be dissected into changes in PA, PL and PLP according to CRP and markers of cellular immunity (KTR and neopterin). PA was positively related to KTR and neopterin but not to CRP, whereas PLP and PL were more strongly associated with CRP than with KTR and neopterin (Ulvik et al., 2014). These observations suggest that during acute phase (reflected by CRP) there is increased uptake of PLP and PL, whereas cellular immune activation is dominated by increased PL degradation to PA. Thus, the PAr index is a measure of both these inflammatory modalities.

Cellular uptake of vitamin B6 includes dephosphoryla­

tion of PLP to PL, which crosses the cell membrane, and is retained within the cells after rephosphorylation to PLP cata­

lyzed by PL kinase (di Salvo et al., 2011). PLP and PL are

interconvertible, and PLP-specific phosphatases exist in most

tissues (Jang et al., 2003). The oxidation of PL to PA is irre­

versible and is believed to be catalyzed by liver Aldehyde oxidase 1 (AOX 1) (Garattini et al., 2009; Merrill et al., 1984), the expression of which is regulated by oxidative stress- related signal pathways (Maeda et al., 2012). PL is also oxidized to PA by aldehyde dehydrogenase (ALDH), which is expressed in many tissues (Stanulovic´ et al., 1976). Thus, expression of both AOX 1 and ALDH is increased during oxi­

dative or aldehyde stress (Maeda et al., 2012; Vasiliou and Nebert, 2005). This may explain the strong association of PAr with the marker of cellular immune activation, KTR, which reflects the activation of indoleamine 2,3-dioxygenase (IDO), an enzyme under redox control and with peroxi­

dase activity (Freewan et al., 2013; Yeung et al., 2015).

The PAr index has recently been shown to be a predic­

tor of incident cancer in the general population. The association was strongest for lung cancer (Zuo et al., 2015).

Inflammation has been assigned a role in lung carcinogen­

esis, which is also predicted by markers of IDO activation and cellular immunity in recent prospective studies (Chuang et al., 2014; Zuo et al., 2014). In patients with angina pec­

toris, PAr was a stronger predictor of all-cause mortality than current smoking, diabetes, hypertension, apolipoproteins or CRP. The association with PAr was strongest in patients with no prior coronary events (Ulvik et al., 2016).

6. Mobilization of vitamin B6 during inflammation

The research reviewed above demonstrates that inflam­

mation leads to a marked reduction in plasma PLP, and small changes in erythrocyte PLP; both plasma and erythrocyte PLP show a minor response to pyridoxine supplementa­

tion, whereas PL increases markedly. These observations suggest that depletion of PLP is confined to certain com­

partments, an idea supported by the results obtained in rats with adjuvant arthritis, which caused a marked reduction in PLP in liver and plasma, but not in muscle (Chiang et al., 2005). Plasma PLP probably reflects the vitamin B6 status in liver (Lumeng et al., 1980), which contains a rapidly ex­

changing PLP pool (Bode and van den Berg, 1991) that is mobilized via circulation to the sites of inflammation. In­

flammation is associated with decreased serum albumin and increased circulating ALP (Aida, 1993; Chiang et al., 2005).

These changes may facilitate mobilization of plasma PLP, by reducing PLP binding to albumin (Bates et al., 1999b; Cheng et al., 2006; Chiang et al., 2005; Huang et al., 2005; Lumeng et al., 1974; Quasim et al., 2005) and, more importantly, by increasing dephosphorylation of free PLP to PL (Narisawa et al., 2001; Vasilaki et al., 2008; Whyte et al., 1985).

Conceivably, altered vitamin B6 distribution during in­

flammation may not be restricted to sites of inflammation, but may also involve unaffected tissues or cells. CRP, is a marker of the IL-1β/TNF-α/IL-6 pathway (Eklund, 2009;

Ridker, 2016). IL-6 upregulates ALP, required for the uptake of PLP from the circulation (Gallo et al., 1997; Kapojos et al., 2003). IL-1β and IL-6 are among the activators of the hypothalamic-pituitary-adrenal (HPA) axis where cortisol is the final key messenger (Kageyama and Suda, 2009; van der Meer et al., 1996; Venihaki et al., 2001). Cortisol has widespread effects in the body and is the main regulator of the physiological stress response including the upregulation

of gluconeogenesis and degradation of protein in muscle, gut, and connective tissue. The liberated amino acids may then be utilized for energy production, synthesis of immunomodulating proteins, immune cell proliferation, and tissue repair. All of these processes require, and may there­

fore increase the cellular demand for PLP. Moreover, an increase in intracellular PLP has been implicated in the mod­

ulation of the cell’s response to glucocorticoids (Tully et al., 1994). Glucocorticoids may have profound effects on vitamin B6 metabolism and distribution as demonstrated in mice given long-term prednisone treatment. Prednisone induced an increase in plasma PLP, PL and PA. There was a concur­

rent increase in the activities of PLP synthesizing enzymes (pyridoxal kinase (PDXK), pyridoxamine 5-phosphate oxidase (PMPO)) and a suppression of pyridoxal 5′-phosphate phos­

phatase (PDXP) in the liver whereas plasma ALP was not affected (Chang et al., 2011).

The changes in tissue vitamin B6 distribution during in­

flammation may modulate PLP-dependent enzymes and metabolic pathways that play a significant role in the in­

flammatory response. Vitamin B6-dependent networks of metabolites with immunomodulating effects and/or that respond to inflammation could be denoted as vitamin B6-dependent inflammatory pathways. Some vitamin B6­

dependent inflammatory pathways that have been extensively studied in recent years are reviewed below.

7. Vitamin B6 and inflammatory pathways 7.1. The kynurenine pathway

The essential amino acid, tryptophan (Trp) is mainly ca­

tabolized along the so-called kynurenine pathway that produces a variety of compounds, collectively termed kynurenines (Fig. 2), many of which have immuno­

modulatory effects. Several enzymes involved in the kyn­

urenine pathway require PLP as co-factor and their expression is regulated by inflammatory cytokines.

7.1.1. The pathway

The first and rate-limiting step in the kynurenine pathway is the oxidative cleavage of the indole ring of Trp to form N-formylkynurenine (Wang et al., 2015), which is enzy­

matically (by formamidase) or spontaneously decomposed to kynurenine (Kyn) and formic acid (Thomas and Stocker, 1999). The formation of N-formylkynurenine is catalyzed by three enzymes, tryptophan 2,3-dioxygenase (TDO), indolamine 2,3-dioxygenase 1 (IDO1) and IDO2 (Ball et al., 2014; Chen and Guillemin, 2009). Kyn is further metabo­

lized to kynurenic acid (KA) or anthranilic acid (AA) through

reactions catalyzed by kynurenine aminotransferases (KAT)

(Han et al., 2010a; Passera et al., 2011; Pinto et al., 2014) or

kynureninase (KYNU) (Phillips, 2014), respectively, both of

which require PLP as co-factor. Alternatively, Kyn is

oxidized to 3-hydroxykynurenine (HK) in a reaction

catalyzed by the FAD-dependent enzyme, kynurenine

3-monooxygenase (KMO) (Smith et al., 2016). HK in turn is

metabolized either to xanthurenic acid (XA) by KAT or

3-hydroxyanthranilic acid (HAA) by KYNU, and HK therefore

occupies a unique position in the kynurenine pathway, since

its further metabolism is dependent on PLP. HAA is

Fig. 2. Kynurenine pathway of tryptophan metabolism and the enzymes involved. The pathway includes a variety of metabolites, collectively termed kynurenines, some of which are neuroactive compounds and some have immunomodulatory effects. The first and rate-limiting step of tryptophan catabolism is the oxidation of tryptophan to N-formylkynurenine, catalyzed by the heme dioxygenases, hepatic tryptophan (2,3)-dioxygenase (TDO; EC 1.13.1.2) and ubiq­

uitous indoleamine (2,3)-dioxygenase (IDO; EC 1.13.11.42); for the latter enzyme two gene homologues, IDO1 and IDO2, have been identified. IDO1 is activated by proinflammatory cytokines such as interferon-γ and tumor necrosis factor-α. N-Formylkynurenine is rapidly converted to kynurenine (Kyn), a reaction that occurs spontaneously or is catalyzed by formamidase (EC 3.5.1.9; not shown). Further metabolism of Kyn involves two pyridoxal 5′-phosphate (PLP)­

dependent enzymes, kynureninase (KYNU; EC 3.7.1.3) and kynurenine aminotransferase (KAT), that catalyze the formation of anthranilic acid (AA) and kynurenic acid (KA), respectively. Alternatively, kynurenine is oxidized in a reaction catalyzed by the FAD-dependent enzyme, kynurenine 3-monooxygenase (KMO; EC 1.14.13.9), to 3-hydroxykynurenine (HK), which is further metabolized to 3-hydroxyanthranilic acid (HAA) or xanthurenic acid (XA) catalyzed by KYNU and KAT, respectively. 3-Hydroxyanthranilate 3,4-dioxygenase (HAAO); EC 1.13.11.6) converts HAA to 2-amino-3-carboxymuconate semialdehyde (ACMS), which spontaneously cyclizes to quinolinic acid (QA). QA is decarboxylated and conjugated with phosphoribosylpyrophosphate to form nicotin­

amide mononucleotide (NAM) in a reaction catalyzed by quinolinate phosphoribosyltransferase (QPRT; EC 2.4.2.19). Modified from Ueland et al. (2015).

oxidized by 3-hydroxyanthranilate 3,4-dioxygenase (HAAO) (Zhang et al., 2005) into 2-amino-3-carboxymuconate semialdehyde (ACMS), which is either enzymatically con­

verted to picolinic acid or spontaneously reassembled to form quinolinic acid (QA). In a reaction catalyzed by quinolinate phosphoribosyltransferase (QPRT), QA is subsequently con­

verted into nicotinic acid mononucleotide (Liu et al., 2007), and ultimately NAD

(Magni et al., 1999). NAD formation is a minor pathway in the liver, and suggested to be regulated by niacin intake (Moffett and Namboodiri, 2003).

7.1.2. Key enzymes

TDO is a heme-containing enzyme that is mainly ex­

pressed in the liver but also in some tumour cells (Opitz et al., 2011; Pilotte et al., 2012; van Baren and Van den Eynde, 2015). It is a high K

enzyme that initiates Trp degradation (Batabyal and Yeh, 2007), is induced by corti­

costeroids and Trp (Salter and Pogson, 1985), and is the main enzyme responsible for catabolism of dietary Trp and main­

tenance of Trp homeostasis (Ball et al., 2014).

IDO1 is also a heme-containing enzyme but with a differ­

ent structure, tissue distribution, regulatory and kinetic properties as compared with TDO (Yeung et al., 2015). It is ex­

pressed extra-hepatically in a variety of cells and tissues, including monocyte-derived macrophages, dendritic cells (Guillemin et al., 2001), epithelial cells, brain (Guillemin et al., 2001, 2005) and cancer cells (van Baren and Van den Eynde, 2015). Notably the morphological features of IDO1-expressing cells often resemble those of antigen presenting cells (Dai and Zhu, 2010). Compared to TDO, IDO1 shows less substrate speci­

ficity and lower K

for Trp (Austin et al., 2009; Pantouris et al., 2014). IDO1 expression is induced by various cytokines, and among these the Th-1-type cytokine, INF-γ, is most impor­

tant (Mandi and Vecsei, 2012). Thus, the ratio between Kyn and Trp (KTR) in the circulation reflects the activation of IDO and serves as a useful marker of cellular Th-1-type immune activation (Schrocksnadel et al., 2006).

IDO2 is a newly discovered paralogue to IDO1 (Murray, 2007) with a similar genomic structure (43% homology) and adjacent chromosomal localization, but with different tissue distribution and kinetic properties (Ball et al., 2007; Metz et al., 2007; Yuasa et al., 2007). IDO2 is constitutively expressed in liver, kidneys, spermatozoa and dendritic cells. It has lower V

, higher K

for Trp and shows less substrate specificity as com­

pared with IDO1 (Ball et al., 2014; Pantouris et al., 2014; Yuasa et al., 2007). Data on its biological role are limited, but IDO2 seems to be important for the induction of several inflamma­

tory cytokines, and may play a role in autoimmune response, immune tolerance and cancer cell surveillance (Metz et al., 2014; Prendergast et al., 2014).

Kynurenine 3-monooxygenase (KMO) is a NAD(P)H­

dependent flavin monooxygenase that catalyzes the conversion of Kyn to HK (Smith et al., 2016). It is a mito­

chondrial enzyme that is expressed at high levels in the liver, kidney and macrophages, but also in microglia in the CNS (Alberati-Giani et al., 1997; Allegri et al., 2003; Guillemin et al., 2003). KMO has low K

in the micromolar range, sug­

gesting that it keeps intracellular Kyn at a low level (Breton et al., 2000). Of note, KMO expression seems to be stimu­

lated during inflammation and immune activation (Campbell et al., 2014; Connor et al., 2008; Parrott and O’Connor, 2015).

Kynurenine aminotransferases (KATs) are PLP-dependent enzymes that exist as four different isoforms, i.e. KATI/II/

III/IV. They have different tissue distribution and different substrate specificities, which is reflected by their alter­

nate names, KATI/glutamine transaminase K/cysteine conjugate β-lyase 1, KATII/aminoadipate aminotransferase (Han et al., 2008, 2009a, 2009b), KATIII/glutamine trans­

aminase L (Yu et al., 2006) and KATIV/mitochondrial aspartate aminotransferase (Han et al., 2010a, 2010b, 2011).

KATI and KATIII have sequence homology and the highest k

/K

values with glutamine (Han et al., 2004, 2009a, 2009b;

Yu et al., 2006). KATII is the most abundant isoform in human brain and is hypothesized to be the main source of cere­

bral synthesis of the neuroprotective kynurenine, KA (Guidetti et al., 2007). Expression of KATI and KATIII, but not KATII, may be down regulated by the proinflammatory cytokine, IL-1β (Zunszain et al., 2012).

Kynureninase (KYNU) is a PLP-dependent enzyme that catalyzes two steps in the kynurenine pathway, i.e. the con­

version of Kyn to AA and HK to HAA (Alberati-Giani et al., 1996a, 1996b; Walsh and Botting, 2002). In humans, there is expression of one KYNU (Alberati-Giani et al., 1996a, 1996b), which is mainly located in cytoplasm and has higher k

/K

for HK than for Kyn (Alberati-Giani et al., 1996b). The formation of HAA is a key step toward the for­

mation of NAD. Notably, KYNU is heavily upregulated during inflammation (Harden et al., 2015) and its expression is in­

creased in murine macrophages by INF-γ (Alberati-Giani et al., 1996a) and in human keratinocytes by IL-17 and TNF-α (Chiricozzi et al., 2011). It has been suggested that KYNU serves as a switch between immunosuppression versus in­

flammation (Harden et al., 2015).

7.1.3. Biological effects from metabolites

Early hypotheses suggested that activation of Trp ca­

tabolism through the kynurenine pathway reduced Trp bioavailability thereby causing Trp starvation and subse­

quently inhibiting growth of pathogens and proliferative cells. Thus, Trp depletion was regarded as antimicrobial and antitumoral defense mechanism induced by INF-γ pro­

duced by immunocompetent cells (de la Maza and Peterson, 1988; Ozaki et al., 1988; Pfefferkorn, 1984). However, recent studies have demonstrated that Kyn is not an inert meta­

bolic intermediate. Kyn has specific roles, such as being an endothelium-derived vasodilator produced during inflam­

mation (Wang et al., 2010), and an endogenous ligand for the aryl hydrocarbon receptor (AhR) (Opitz et al., 2011), a ligand sensitive transcription factor that mediates (among other functions) immunosuppression (Bessede et al., 2014).

Activation of AhR in turn may upregulate IDO (Bessede et al., 2014) thereby promoting the generation of regulatory T cells (Mezrich et al., 2010).

Kynurenic acid (KA), an end product of the KAT branch of the kynurenine pathway, is a neuroprotective and immunomodulating metabolite that inhibits the N-methyl­

D-aspartate (NMDA) receptor and α-7-nicotine acetylcholine receptor (α-7-NAChR) in the CNS (Moroni et al., 2012).

Notably, KA is an endogenous ligand of the orphan G protein­

coupled receptor 35 (GPR35) (Wang et al., 2006) and AhR

(DiNatale et al., 2010), which may in part mediate its

immunomodulating effects. KA has also been considered as

a potent endogenous antioxidant that scavenges hydroxyl radicals, superoxide anions and peroxynitrite (Lugo-Huitrón et al., 2011).

Anthranilic acid (AA) is generally considered a biologi­

cally inert metabolite with possible hydroxyl radical scavenging properties (Miche et al., 1997). The redox active metabolite, 3-hydroxyanthranilic acid (HAA), is a potent an­

tioxidant (Giles et al., 2003; Leipnitz et al., 2007; Thomas et al., 1996) with anti-inflammatory and immunosuppres­

sive effects (López et al., 2008; Weber et al., 2006) involving the induction of apoptosis in activated T-cells (Fallarino et al., 2002; Hayashi et al., 2007; Lee et al., 2010b). Possible mecha­

nisms include increased heme oxygenase-1 (HMOX1) expression (Krause et al., 2011), inhibition of inducible nitric oxide synthase (iNOS) (Sekkaï et al., 1997), depletion of glu­

tathione (GSH) (Lee et al., 2010b), and inhibition of 3-phosphoinositide-dependent protein kinase 1 (PDK1) phosphorylation (Hayashi et al., 2007).

3-Hydroxykynurenine (HK) is a redox active, reactive oxygen species (ROS) generating Trp metabolite (Vazquez et al., 2000) with neurotoxic effects (Smith et al., 2009).

Recent studies demonstrated, however, that HK has a dual role by serving either as a prooxidant or an antioxidant (Leipnitz et al., 2007), depending on the experimental system, suggesting that HK is not neurotoxic or cytotoxic

ing the redox defense system, including increasing the expression of Nuclear factor erythroid 2-related factor 2 (Nrf2) (Colín-González et al., 2014a, 2014b), the master me­

diator of anti-inflammatory effects (Ma, 2013).

Xanthurenic acid (XA) is formed by transamination of HK, which has been considered to represent a reaction pre­

venting build-up of high, potentially toxic concentration of HK (Gobaille et al., 2008). XA crosses the blood–brain barrier, and may play a role in neurotransmission/neuromodulation (Gobaille et al., 2008) by activating glutamate receptor mGlu2/3 (Fazio et al., 2015). It is a metal-chelating (Murakami et al., 2006) and photochemically active (Roberts, 2001) compound with antioxidant (Christen et al., 1990) and prooxidant (Murakami et al., 2006) properties, depending on the experimental system. These properties have been related to its possible role in cataractogenesis (Roberts, 2001), lens epithelial cell apoptosis (Malina et al., 2002), and de­

velopment of diabetes (Oxenkrug, 2015). Notably, XA has recently been found to be a potent inhibitor of the biosyn­

thesis of tetrahydrobiopterin (Haruki et al., 2015), which serves as a cofactor of three aromatic amino acid hydroxy­

lases and nitric oxide synthase (NOS).

3-Hydroxyanthranilic acid dioxygenase (HAAO) con­

verts HAA to 2-amino 3-carboxymuconate semialdehyde (ACMS), a fraction of which is spontaneously converted to quinolinic acid (QA) (Malherbe et al., 1994). The most prom­

inent feature of QA is its ability to serve as an endogenous NMDA-type glutaminergic receptor agonist, which ex­

plains its excitotoxicity, cytotoxic effects on neurons and astrocytes, and induction of seizures. Additional biological effects of QA include increased neuronal glutamate release, induction of oxidative stress (Colín-González et al., 2014a, 2014b; Vandresen-Filho et al., 2015), disruption of the blood–

brain barrier, induction of neuronal nitric oxide synthase (nNOS) and iNOS (Braidy et al., 2009), cytoskeleton

destabilization (Pierozan et al., 2014), increased tau phos­

phorylation (Rahman et al., 2009) and increased expression of several proinflammatory cytokines (Lugo-Huitrón et al., 2013). Human neurons do not produce significant amounts of QA, but neuronal cells take up QA produced by acti­

vated microglia and infiltrating macrophages (Guillemin, 2012). The kynurenine pathway and thereby the synthesis of QA by these cells is substantially increased during in­

flammation, leading to accumulation of QA, which explains the key role of QA in neuroinflammation (Guillemin, 2012).

7.1.4. Kynurenine pathway metabolites and vitamin B6 status

It has been consistently demonstrated that the concen­

trations of kynurenines in urine and plasma/serum are affected by vitamin B6 status in humans. Vitamin B6 defi­

ciency caused a more than 30-fold increase in urinary excretion of XA, Kyn and HK after a tryptophan load in women. There was also a moderate increase in urinary HAA and QA (Yeh and Brown, 1977). Supplementation of sub­

jects having adequate vitamin B6 status with PN decreased urinary excretion of XA, Kyn and HK after a tryptophan load, suggesting that PLP-dependent enzymes of the kynuren­

ine pathway may not be fully saturated with the PLP cofactor (Leklem, 1971). The responsive metabolites, XA, Kyn and HK, precede the cleavage of HK to HAA catalyzed by KYNU, which is more sensitive than KAT to vitamin B6 depletion (Ogasawara et al., 1962; van de Kamp and Smolen, 1995).

Among plasma kynurenines, only HK shows a marked in­

crease in subjects with low plasma PLP (Theofylaktopoulou et al., 2014) and is reduced in subjects supplemented with PN (Midttun et al., 2011). Plasma KA, AA, XA and HAA show a positive relation with PLP (Theofylaktopoulou et al., 2014;

Ulvik et al., 2013), while KA and HAA show a slight de­

crease following PN supplementation (Midttun et al., 2011).

These variations in urinary and plasma kynurenines ac­

cording to vitamin B6 status reflect the critical role of PLP availability for steady state metabolite concentrations and conceivably flux through the kynurenine pathway.

Urinary excretion of XA after a tryptophan load and the plasma HK/XA ratio have been established as functional markers of vitamin B6 status, as summarized in a recent review article (Ueland et al., 2015).

7.1.5. Kynurenine pathway metabolites and chronic diseases

The kynurenine pathway is upregulated under patho­

logical conditions characterized by involvement of immune activation in pathogeneses or established disease. Such ac­

tivation causes Trp depletion and formation of neuroactive kynurenines with immunomodulating effects, and has been suggested to play a role in numerous disorders, including neurodegenerative diseases (Parrott and O’Connor, 2015), depression (Meier et al., 2015; Réus et al., 2015), schizo- phrenia, infections, osteoporosis (Michalowska et al., 2015), rheumatoid arthritis, cancer and cardiovascular disease (Chen and Guillemin, 2009; Munn and Mellor, 2013).

In neurodegenerative diseases, including Alzheimer’s disease, Parkinson’s disease and Huntington’s disease, there appears to be a shift toward QA and HK synthesis, and away from KA production (Maddison and Giorgini, 2015). Mul­

tiple sclerosis relapse (Rajda et al., 2015) and schizophrenia

(Parrott and O’Connor, 2015) have been associated with in­

creased KA in plasma and CSF. Increased concentrations of key metabolites of the kynurenine pathway are linked to processes associated with development of cardiovascular diseases, such as immune activation, inflammation, gener­

ation of ROS, endothelial dysfunction, and vasodilation (Wang et al., 2015). Both IDO1 and less frequently TDO are expressed in human tumours, which explain increased Trp catabolism, Trp depletion and increased levels of immunomodulating kynurenines in cancer. These are mecha­

nisms that mediate tumoral immune resistance and suppression of cancer immunosurveillance (van Baren and Van den Eynde, 2015).

Most investigations of the kynurenine pathway in rela­

tion to disease involve experimental studies in animals or cells and small clinical studies in humans. Recently, large cross-sectional and prospective studies including thou­

sands of participants have been published. KTR (which reflects IDO activation), Kyn (Sulo et al., 2013), HK, and to a lesser extent other kynurenines (Eussen et al., 2015;

Pedersen et al., 2015; Zuo et al., 2016) in plasma, and KTR in urine (Pedersen et al., 2013) are associated with risk of cardiovascular disease. Additionally, KTR is associated with risk of overall cancer, with strongest risk estimates ob­

served for lung cancer (Chuang et al., 2014; Zuo et al., 2014), a malignancy where inflammation is believed to play a role in carcinogenesis. KTR and/or some downstream kynurenines were also associated with bone mineral density and hip fracture in a recent large cross sectional, popula­

tion based study (Apalset et al., 2014a, 2014b).

7.2. Sphingolipids

The vitamin B6 antagonist, 4′-deoxypyridoxine (DOP) (Trakatellis et al., 1997), and the food colorant, 2-acetyl-4­

tetrahydroxybutylimidazole (THI) (Gobin and Paine, 1989;

Ohtoyo et al., 2015), have been shown to cause a reduc­

tion in circulating lymphocytes when given to rodents. These effects were fully reversible by addition of vitamin B6 (Gobin and Paine, 1989; Trakatellis et al., 1997). In 2005, the group of Jason Cyster reported that DOP and THI inhibit the enzyme, sphingosine 1-phosphate lyase (SPL), thereby causing a build-up of sphingosine 1-phosphate (S1P) in lym­

phoid organs, and concomitant lymphopenia, the latter reflecting inhibition of lymphocyte egress into the circula­

tion (Schwab et al., 2005).

S1P is a bioactive sphingolipid, the

which starts with the PLP-dependent enzyme, serine palmitoyltransferase (SPT). This is the rate limiting reac­

tion forming 3-ketosphinganine, which is converted to ceramide (a sphingolipid) in a series of reactions (Bourquin et al., 2011). Alternatively, ceramide is formed from dietary sphingomyelin through the action of sphingomyelinase.

Ceramidase (CA) catalyzes the conversion of ceramide to sphingosine, which is phosphorylated to S1P by two sphingokinases (SKs), SK1 and SK2 (Liu et al., 2012; Pyne et al., 2015). S1P can be dephosphorylated to sphingosine by two S1P-specific phosphatases, SPP1 and SPP2, or by lipid non-specific phosphatases (Liu et al., 2012). Notably, the ir­

reversible degradation of S1P to trans-2-hexadecenal and phosphoethanolamine is catalyzed by the PLP-dependent

sphingosine 1-phosphate lyase (SPL), which is an enzyme that regulates the steady state S1P concentration in tissues and the circulation (Aguilar and Saba, 2012; Bourquin et al., 2011) (Fig. 3).

S1P formed in the intracellular compartment is se­

creted into the extracellular environment; S1P concentration is relatively low in interstitial fluid and tissues (~nM), but high in the lymph (~0.1 μM) and the bloodstream (~1 μM), where it circulates bound to high-density lipoprotein (Blaho et al., 2015) and albumin (Ksia˛z˙ek et al., 2015; Yatomi, 2008).

The biological effects of S1P appear to be mediated primar­

ily by S1P serving as a high-affinity ligand of five specific cell-surface G-protein-coupled receptors (GPCRs), desig­

nated S1P

(Blaho and Hla, 2014). In addition, S1P activates Nuclear factor-κB (NF-κB) and Signal transducer and acti­

vator of transcription 3 (STAT3), two transcriptional regulators that serve as master switches in inflammation and carcinogenesis (Alvarez et al., 2010; Lee et al., 2010a).

The receptors S1P

mediate effects in many facets of mammalian biology, including integrity and development of the vasculature and nervous system (Blaho and Hla, 2014;

Proia and Hla, 2015). Activation of the S1P

is critical for the egress of lymphocytes residing in the secondary lymphoid organs and thymus (Chi, 2011; Cyster and Schwab, 2012;

Proia and Hla, 2015) and is involved in regulating differ­

entiation of T-cells, including T helper 17 (Garris et al., 2013) and T helper 1/regulatory T cell balance (Liu et al., 2010).

Ceramide is formed

synthesis of S1P. The PLP-dependent enzyme, SPT, which links ceramide to vitamin B6 function, catalyzes the first step.

Enzymes involved in other routes of ceramide synthesis or degradation are not vitamin B6-dependent. Ceramide kinase (CK) converts ceramide to ceramide 1-phosphate (C1P) (Fig. 3). Ceramide and C1P have been liked to inflamma­

tion, but data are less compelling than for S1P, and a surface receptor for C1P is yet to be validated. (Maceyka and Spiegel, 2014).

C1P activates cytosolic phospholipase–A2α and thereby synthesis of arachidonate for the production of eicosanoids. C1P also plays a role in the processing of the proinflammatory cytokine TNF-α. Other functions of C1P include promoting cellular proliferation and growth, macrophage migration, and inhibition of apoptosis (Gomez-Muñoz et al., 2015).

7.3. Transsulfuration pathway and hydrogen sulfide formation

The gaseous messenger hydrogen

(H

S) has emerged as a regulator of inflammatory response. H

S has antiinflammatory effects at low, physiological concentra­

tions, but is proinflammatory at high concentrations (Bhatia, 2012; Gemici and Wallace, 2015; Whiteman and Winyard, 2011). H

S also serves as a regulator of numerous other phys­

iological functions, including vasodilation, angiogenesis (Liu et al., 2011), neurotransmission, apoptosis and insulin release (Bełtowski, 2015; Paul and Snyder, 2012). Cysteine is the major thiol in plasma, and the redox state of cysteine/

cystine may itself play a role in inflammatory signaling (Go

and Jones, 2011). It also serves as a key component in GSH

synthesis (Stipanuk and Ueki, 2011). GSH is not only an

Fig. 3. Metabolism of sphingolipids and the enzymes involved. Sphingolipids are synthesized de novo in the endoplasmic reticulum and synthesis is ini­

tiated by the condensation of L-serine with palmitoyl-CoA catalyzed by the PLP-dependent enzyme serine palmitoyltransferase (SPT; EC 2.3.1.50) forming 3-ketosphinganine, which is metabolized to ceramide through three enzymatic steps. Newly synthesized ceramide is transported to the Golgi apparatus where it is converted to sphingomyelin in a reaction catalyzed by sphingomyelin synthase (SMS; EC 2.7.8.27). Degradation in the lysosomes and plasma membrane of higher-order sphingolipids leads to the formation of ceramide, which is converted to sphingosine in a reaction catalyzed by ceramidase (CA;

EC 3.5.1.23). Sphingosine kinase(s) (SKs; EC 2.7.1.91) catalyzes the conversion of sphinogosine to sphinogosine 1-phosphate (S1P) in multiple compart­

ments. S1P can be dephosphorylated to sphinogosine by enzymes like the ecto enzymes, lipid non-specific phosphatases (LPP1, LPP2 and LPP3), and by S1P-specific phosphatases (SPP1 and SPP2; EC 3.1.3.-), localized in the plasma membrane and endoplasmic reticulum, respectively. Notably S1P is irre­

versibly degraded in endoplasmic reticulum by PLP-dependent enzyme S1P lyase (SPL; EC 4.1.2.27), which is the key regulator of the steady state S1P concentration in tissues and the circulation. Abbreviations: SMAs, sphingomyelinase(s); CS, Ceramide synthase; CK, Ceramide kinase; CPP, Ceramide 1-phospahte phosphatase.

antioxidant, detoxifying reactive oxygen species, but also a 2009). The sequential action of these two enzymes com­

signaling molecule that regulates innate immunity and in- prises the transsulfuration pathway responsible for the flammation (Ghezzi, 2011), effects partly mediated by conversion of homocysteine to cysteine through the inter­

cytokines and redox-sensitive transcription factors like mediate, cystathionine (Stipanuk and Ueki, 2011) (Fig. 4).

NF-κB and Hypoxia-inducible factor-1α (HIF-1α) (Haddad About 20–50% of cysteine used for GSH synthesis in the liver

and Harb, 2005). is derived from the transsulfuration pathway (Gregory et al.,

The formation of H

S and cysteine involves two PLP- 2016; Mosharov et al., 2000), which seems to be respon­

dependent enzymes, cystathionine β-synthase (CBS) (Miles sible for the bulk of H

S generated for regulatory purposes

and Kraus, 2004) and cystathionine γ-lyase (CSE) (Kraus et al., (Kabil and Banerjee, 2014).

Fig. 4. The transsulfuration pathway, H2S production and the enzymes involved. Methionine is activated to the universal methyl donor, S-adenosylmethionine, which is converted to S-adenosylhomocysteine during S-adenosylmethionine-dependent transmethylation reactions. S-adenosylhomocysteine in turn is hydrolyzed to adenosine and homocysteine in a reaction catalyzed by S-adenosylhomocysteine hydrolase (not shown). Homocysteine is either remethylated to methionine (not shown) or converted to cysteine via the transsulfuration pathway, where homocysteine is converted to cysteine through the sequen­

tial action of two PLP-dependent enzymes, cystathionine β-synthase (CBS; EC 4.2.1.22) and cystathionine γ-lyase (CSE; EC 4.4.1.1). H₂S is produced through seven non-canonical reactions catalyzed by CBS and CSE, as indicated.

Both CBS and CSE are ubiquitous, but CBS is expressed pre­

dominately in the brain and kidney while CSE is expressed primarily in the peripheral tissues like liver and cardiovascu­

lar tissue. In the first step in the transsulfuration pathway CBS catalyses the synthesis of cystathionine by condensing ho­

mocysteine with serine. H

S is formed during non-canonical reactions catalyzed by CBS. These reactions include cysteine desulfuration by linking homocysteine with cysteine (instead of serine) (Chen et al., 2004), and conversions of cysteine to serine or lanthionine (Stipanuk and Ueki, 2011).

CBS has a catalytic domain that binds PLP and re­

sembles the catalytic core of other members of fold II family of PLP-dependent enzymes. It has two regulatory domains at the N- and C-terminals, respectively. The N-terminal heme domain serves as a redox gas sensor mediating increased CBS activity under oxidizing condi­

tions. Interaction of carbon monoxide (CO) and nitric oxide (NO) with the ferrous (Fe

) form of this regulatory site reduces enzyme activity whereas air oxidation leads to recovery of enzyme activity; thus CBS serves as a point of cross-talk between modulators of inflammation, like CO, NO and H

S. The C-terminal regulatory domain inhibits enzyme activity of the full-length enzyme, and the inhibi­

tion is relieved by interaction of S-adenosylmethionine (SAM) with an allosteric site. SAM thereby coordinates the balance between methylation and redox-balance (Kabil and Banerjee, 2014).

CSE catalyzes the second step in the transsulfuration pathway where cystathionine is cleaved to cysteine and α-ketobutyrate. H

S production results from side reac­

tions of CSE with cysteine, homocysteine or both as substrates (Zhao et al., 2014) (Fig. 4).

CSE appears to be more inducible than CBS (Zhao et al., 2014). Numerous compounds affect H

S production and/

or regulate CSE at the transcriptional, post-transcriptional and/or post-translational levels. These include inflamma­

tion mediated by TNF-α, the transcription factors Specificity protein-1 (Sp1), oxidative stress, microRNAs (miR-21 and miR-30), testosterone, estrogens, dexamethasone, phenyl­

ephrine, insulin, glucose, NO, CO, and calcium-calmodulin (Zhao et al., 2014). Notably, calcium may interact with PLP to regulate CSE activity, and H

S production at low calcium levels greatly depends on PLP (Mikami et al., 2013).

PLP serves as a cofactor for both CBS and CSE in the side reactions producing H

S (Gregory et al., 2016). Kinetic simulations suggest that CBS accounts for up to 70% of transsulfuration derived H

S formation, and the condensa­

tion of homocysteine and cysteine seems to create more H

S than condensation of two molecules of cysteine (Singh et al., 2009). The CSE-catalyzed condensation of two homocys­

teine molecules may be less efficient under normal conditions, but may increase during hyperhomocysteinemia.

Since CSE is more sensitive than CBS to available PLP, it has been suggested CSE-catalyzed formation of H

S is most ex­

tensively affected by vitamin B6 status (Gregory et al., 2016).

The transsulfuration pathway is influenced by vitamin B6 status as demonstrated by increased plasma cystathionine in humans (da Silva et al., 2013; Davis et al., 2006; Lamers, 2011;

Midttun et al., 2007) and rats (Stabler et al., 1997) during moderate vitamin B6 deficiency. In rats there is a concurrent increase in cystathionine in liver and muscle (Lima et al., 2006;

Swendseid et al., 1964). Studies with isolated cells demon­

strated higher cystathionine when cultured at moderate vitamin B6 deficiency than at severe deficiency (da Silva et al., 2014). CBS and CSE have similar affinities for PLP (Gregory et al., 2016); therefore build-up of cystathionine has been ex­

plained by higher rate of turnover for CSE than for CBS and by loss of CSE apoenzyme at low PLP (Lima et al., 2006). Build­

up of cystathionine during vitamin B6 deficiency may actually maintain cysteine flux, which in turn provides sufficient cys- teine to prevent a decline in GSH (Davis et al., 2006; Lima et al., 2006). This explains why vitamin B6-dependent changes in transsulfuration do not decrease GSH content in the liver during mild- to moderate vitamin B6 deficiency (Lima et al., 2006).

On the contrary, dietary vitamin B6 restriction causes eleva­

tion of GSH in human plasma (Davis et al., 2006) and rat liver (Lima et al., 2006). This apparent paradox has been explained by a GSH response to oxidative stress induced by vitamin B6 restriction (Nijhout et al., 2009).

Studies with isolated cells demonstrated impaired syn­

thesis of H

S and the H

S biomarkers, homolanthionine and lanthionine, during vitamin B6 restriction (DeRatt et al., 2014). However, for unknown reasons, moderate, short- term vitamin B6 insufficiency did not affect plasma concentrations of homolanthionine and lanthionine in healthy subjects (DeRatt et al., 2016).

The mechanisms whereby the products of the transsulfuration enzymes, H

S and cysteine, modulate in­

flammatory response are complex (Whiteman and Winyard, 2011). Recently, there has been a focus on H

S signaling by sulfhydration (persulfidation), which involves modifica­

tion of cysteine group of numerous target proteins by conversion of a –SH group to a more reactive –SSH group (Paul and Snyder, 2015). H

S may modulate inflammatory responses through sulfhydration of the transcription factor, NF-κB and sulfhydration of a protein that sequesters Nrf2 (Paul and Snyder, 2015).

7.4. Serine and glycine

The biosynthesis and metabolism of serine and glycine are closely linked, and they provide precursors for the syn­

thesis of proteins, nucleic acids and lipids, which are required for the proliferation of immune cells (de Koning et al., 2003;

Wang et al., 2013). Serine is a component in the synthesis of the lipid mediator, S1P (Bourquin et al., 2011), whereas glycine is involved in the synthesis of GSH (Stipanuk and Ueki, 2011). In addition, the role of glycine as an immune modulator with anti-inflammatory effects (Wheeler et al., 1999; Zhong et al., 2003) has been consistently demon­

strated in isolated cells (Blancas-Flores et al., 2012;

Garcia-Macedo et al., 2008), experimental animals (Alarcon-Aguilar et al., 2008; Almanza-Perez et al., 2010;

Gundersen et al., 2007; Takahashi et al., 2008; Vieira et al., 2015), and in humans (Cruz et al., 2008; Zhong et al., 2003).

Suggested mechanisms include activation of the non­

neuronal glycine-gated chloride channel (Froh et al., 2002;

Wheeler et al., 1999), reduction of TNF-α and IL-1β (Hartog et al., 2007), and increased levels of the anti-inflammatory cytokine, IL-10 (Bruck et al., 2003).

Both serine and glycine in plasma have high within­

subjects reproducibility over time (Cope et al., 2013; Midttun

et al., 2014). This observation of high individuality suggests that plasma concentrations are tightly regulated, but also ensures reliable estimates of biomarker status over time from single point measurements in epidemiological and clinical studies. Studies in humans have demonstrated that high glycine is associated with low risk of cardiovascular disease (Ding et al., 2015), type 2 diabetes (Floegel et al., 2013; Klein and Shearer, 2016), impaired insulin sensitivity and low β-cell secretory ca­

pacity (Palmer et al., 2015), observations that are in agreement with its role as an antiinflammatory agent.

The mitochondrial glycine cleavage system (GCS) and cy­

toplasmic and mitochondrial serine hydroxymethyltransferase (cSHMT and mSHMT) are PLP-dependent enzymes involved in serine and glycine metabolism (Scheer et al., 2005; Tibbetts and Appling, 2010) (Fig. 5). GCS catalyzes the decarboxyl­

ation of glycine to CO

and NH

, whereas SHMT is responsible for the interconversion of serine and glycine; both reactions are linked to the formation of methylenetetrahydrofolate that provides activated C1-units for the synthesis of purines, thymidylate and methionine. Serine and glycine are quanti­

tatively the most important C1-donors in humans (Appling, 1991; Tibbetts and Appling, 2010).

Serine and glycine metabolism is influenced by vitamin B6 status, as demonstrated by studies in human (Lamers et al., 2009), experimental animals (Scheer et al., 2005) and isolated cells (da Silva et al., 2014). In humans, plasma glycine and serine were increased after one to two weeks of vitamin B6 depletion; both glycine and serine were normalized after PN supplementation (Park and Linkswiler, 1971). Stable isotope flux studies in humans have demonstrated in­

creased plasma glycine and to a lesser extent serine after

28-day controlled vitamin B6 restriction resulting in mod­

erate vitamin B6 deficiency (da Silva et al., 2013; Davis et al., 2005; Lamers et al., 2009; Nijhout et al., 2009). Concentra­

tions of glycine in plasma, muscle (Swendseid et al., 1964), and liver (Runyan and Gershoff, 1969; Scheer et al., 2005) were increased in rats fed a vitamin B6-deficient diet, and in HepG2 cells cultured in a medium with low vitamin B6 (da Silva et al., 2014).

Low SHMT in rat liver has been reported during vitamin B6 deficiency (Martinez et al., 2000), but results from mathematical modeling suggest that reduced glycine de­

carboxylase is the main cause of glycine accumulation (Lamers et al., 2009; Nijhout et al., 2009), which in turn results in more glycine being converted to excess serine by SHMT (Nijhout et al., 2009). Notably, the decarboxylase flux appears to be only slightly reduced during moderate vitamin B6 deficiency, which has been explained by maintained flux through elevated glycine concentration and a high K

for glycine of the decarboxylase reaction (Fujiwara and Motokawa, 1983; Lamers et al., 2009; Nijhout et al., 2009).

8. Conclusion

Published results demonstrate that inflammation, immunoactivation and related diseases are associated with up to 50% reduction in plasma PLP, and minor changes in erythrocyte PLP and functional vitamin B6 biomarkers. Low plasma PLP parallels reduction in liver PLP, whereas vitamin B6 in muscle is not affected. Vitamin B6 intake or supple­

mentation improves some immunological parameters in vitamin B6-deficient animals and humans. The available

Fig. 5. Serine hydroxymethyltransferase (SHMT) and the glycine cleavage system (GCS). SHMT (EC 2.2.2.1) is a PLP–dependent enzyme that catalyzes the reversible conversions of serine to glycine and tetrahydrofolate to 5,10-methylenetetrahydofolate. In mammals there are two isoforms, a cytoplasmic (cSHMT) and a mitochondrial (mSHMT) form. GCS is also called glycine decarboxylase complex or, when run in the reverse direction, glycine synthase. GCS is a mitochondrial multienzyme system that is composed of four individual proteins, three specific components (P-, T-, and H-proteins), and one housekeep­

ing enzyme, dihydrolipoamide dehydrogenase (L-protein). P-protein is a PLP-dependent glycine decarboxylase (glycine:lipoylprotein oxidoreductase;

EC 1.4.4.2). This system is triggered by high glycine and catalyzes the oxidative cleavage of glycine. When coupled to SHMT, the overall reaction becomes:

2 glycine + NAD⁺+ H2O --> serine + CO2 + NH3 + NADH + H⁺ . Modified from Ueland et al. (2015).